Incident guide

AI tells shoppers your product is out of stock. It is not.

A shopper asks an AI assistant about your product and gets told it is sold out, discontinued, or costs a different price than it does. They do not click through to check. They buy elsewhere, and you never see the lost sale. Wrong availability and price answers are among the most damaging AI mistakes for a store because they end the purchase journey silently. Here is where they come from and how to fix them.

First: confirm what the AI actually claims

Ask the engine about the product several ways, several times: "how much does [product] from [brand] cost", "is [product] in stock at [brand]", "where can I buy [product]". AI answers vary between asks, so a single wrong answer proves little. If the wrong price or stock claim repeats across phrasings, note which engine says it and, where shown, which sources it cites. Those citations usually contain the culprit.

Where wrong stock and price answers come from

Almost always, the engine is faithfully repeating a stale or conflicting source rather than inventing the claim. The usual suspects: an old crawl of your product page from a period when the item really was sold out, a third-party marketplace or price-comparison listing that no longer matches your store, structured data on your own page that disagrees with the visible text, or a variant mix-up where one sold-out size is generalized to the whole product.

The fixes that work, in order of leverage

  1. 1.Fix your product structured data

    Your product page's schema.org markup should state offers with the current price, priceCurrency, and availability set to InStock. Retrieval engines read structured data first, and a page whose markup says OutOfStock will keep generating wrong answers no matter what the visible page says. On Shopify, verify your theme emits correct availability per variant.

  2. 2.State price and availability in visible text

    Put the price and "in stock, ships in N days" in plain page text near the top of the product page, not only in widgets rendered by scripts. Engines quote what they can read; some crawlers execute little or no JavaScript.

  3. 3.Correct the cited third-party sources

    If the engine cites a marketplace listing, a price-comparison site, or an old article, that page is the problem. Update or delist stale marketplace listings and contact comparison sites with the current data. Third-party corrections typically take days to two weeks to be re-crawled.

  4. 4.Keep your product feeds in sync

    Google Merchant Center and marketplace feeds are high-authority sources for price and availability. A feed that lags your catalog teaches every downstream engine the wrong numbers. Automate the sync and audit it after price changes.

  5. 5.Monitor with truth-checks against your live catalog

    The only way to know an engine is misquoting you is to ask it and compare against your real data. ShopBird does this weekly: it asks the engines shopper questions about your products and verifies every price and stock claim against your live Shopify product data, flagging discrepancies with the exact answer that got it wrong.

Honest expectations, engine by engine

Perplexity re-crawls cited pages within days, so own-site fixes land fastest there. ChatGPT with browsing follows in days to weeks. Google AI results depend on Google's index and its shopping feed pipeline, where Merchant Center accuracy matters most. Answers from training data cannot be corrected directly and persist until the next model update, which is one more reason to keep a timestamped record of every wrong answer: if a misquote causes measurable harm, the record is your evidence. And the report buttons on the answers are feedback, not overrides; use them, but do not wait on them.

Frequently asked questions

Why does ChatGPT think my product is out of stock?

It is usually repeating a stale source: an old crawl of your product page, a lagging marketplace or price-comparison listing, or structured data on your own page that says OutOfStock while the visible page says otherwise. Engines repeat sources; find the stale one and fix it.

Will fixing my schema.org markup fix the AI answer?

For engines that retrieve your page (Perplexity, browsing ChatGPT, Google), correct availability and price markup is the highest-leverage fix and typically propagates in days to weeks. It does not help answers generated purely from training data, which change only at the next model update.

The AI quotes an old sale price. Is that fixed the same way?

Yes. Wrong price answers follow the same path as wrong stock answers: correct the structured data, the visible page text, product feeds, and any cited third-party listings. Price claims are also the easiest to prove wrong, so keep timestamped captures of the bad answers.

How fast do wrong stock answers go away after I fix the source?

Days on Perplexity, days to weeks on browsing ChatGPT, weeks on Google AI results, and not until the next model update for training-data answers. Re-check weekly rather than assuming one fix landed everywhere.

How do I catch AI misquoting my prices before shoppers do?

Continuous verification. ShopBird asks the major AI engines real shopper questions about your products every week and truth-checks every price and stock claim against your live Shopify product data. Discrepancies raise an alert with the hash-stamped answer as evidence. The free shopping scan gives you a one-off check right now.

Find out who AI sends your shoppers to, before it costs sales

Run the free AI shopping scan right now, or let ShopBird shop your store weekly across five AI engines: share of recommendation against named competitors, every price and stock claim truth-checked against your live Shopify product data, and a hash-stamped record of every answer.

Run the free shopping scan→Get the Shopify app ↗

New to ShopBird? See how the AI mystery shopper works →